The New Technique That Finds All Known Human Viruses In Your Blood

schwit1 writes with this story at the Atlantic that profiles Ian Lipkin and his new method for quickly detecting all known human viruses in a sample: Ian Lipkin, a virus hunter from Columbia University, recently received a blood sample from colleagues at the National Institutes of Health. They came from a man who had received a bone-marrow transplant and had fallen mysteriously ill, with evidence of severely inflamed blood vessels. In analyzing a similar case a few years back, Lipkin had discovered a new polyomavirus, part of a family that can cause disease in people with compromised immune systems. Perhaps this new case would yield another new virus. It didn't. Instead, when Lipkin's team ran the sample through a system that they had devised to detect human viruses, they found that the man was infected with dengue virus. In hindsight, that made sense-he had recently returned from Vietnam, where dengue is prevalent. But the thing is: The team wasn't looking for dengue virus.

"It wasn't what we anticipated, but we didn't have to make a priori decisions about what we planned to find," Lipkin says. "When people analyze samples from people who are ill, they have some idea in mind. This is probably an enterovirus, or maybe it's a herpesvirues. They then do a specific assay for that particular agent. They don't usually have the capacity to look broadly." The new system, known as VirCapSeq-VERT, barrels past this limitation. Lipkin, together with fellow Columbia professors Thomas Briese and Amit Kapoor, designed it to detect all known human viruses, quickly, efficiently, and sensitively. By searching for thousands, perhaps millions, of viruses at once, it should take a lot of the (educated) guesswork out of viral diagnosis.

Comments

Slightly more technical

[Fwipp]

In slightly more technical terms, they've designed a system that selectively targets & amplifies ~2 million DNA sites; chosen from the genomes of all known infectious viruses. The scientists basically apply this assay to the infected cells (I'm assuming they take a blood sample or something), leaving them with DNA that matches those targets. Then, they run those DNA fragments through a sequencer, and see what they got. From there, they can deduce which virus was present in the original sample.

Re:Slightly more technical

[Immerman]

You're ignoring the second half of their suggestion - test for the "detected" viruses that could cause the symptoms. If you had a 2% false positive rate, and 10% of the "viral library" your checking against could cause the symptoms, then you're averaging 2%*10%*700 = 1.4 false positives per patient.

Of course RTFAing tells me that they're expecting no false positives at all thanks to the fact that this technique provides the complete genome of everything detected, as it occurs in the patient. It's not a "yes/no" like most current tests, or even a "70% certainty", it tells you that "this *exact* genome was in the sample", even if it's something that you've never seen before that was just similar enough to stick to one of the "hooks". You then look up the freshly-sequenced genome in your library to figure out what it is. False positives would imply that the replication system somehow managed to spontaneously create recognizable genomes from scratch.

As I read it the methodology is:
1) catch a wide range of "suspicious" viral DNA
2) replicate it so that you can...
3) sequence it all
4) compare sequenced DNA to database to identify pathogens and unknown viruses.

(1) will likely generate a fair amount of false positives, but unlike traditional tests at that point it's just bringing in "suspicious characters" for questioning. (2) and (3) should be pretty much immune to false positives, and (4) will give you a fully "fingerprinted" list of "confirmed-present" viruses, along with a confidence-rated identification of anything with a match in the database.

Re:Slightly more technical

[AmiMoJo]

A few false positives would be fine, you can just run other more traditional tests to rule them out. This sort of diagnostic is for people who have problems that cannot be readily diagnosed. The choice is either do many expensive tests for different and very unlikely viruses, or use this technique to narrow it down to a small field.

You wouldn't start treatment on the results of this test alone, you would confirm them first.

Re:application of "whole proteome tiling microarra

[Michael Woodhams]

My understanding from a very quick skim of the paper (open access, here) is that they are not using microarrays. They have a mixture of a very large (2 million) number of probes to match DNA/RNA sequences of all known viruses which infect vertebrates. They use these to amplify viral sequences and then use normal high throughput DNA sequencing (Illumina, in this case) to see what they've got. They claim that it is sensitive to both DNA and RNA viruses (and all the variations - double, single stranded etc.) Being able to detect both DNA and RNA in a single test mildly surprises me, but I'm only slightly familiar with DNA sequencing technology, so maybe it isn't a big deal.

They do say "A biotinylated oligonucleotide library was synthesized on the NimbleGen cleavable array platform and used for solution-based capture of viral nucleic acids present in complex samples containing variable proportions of viral and host nucleic acids." Perhaps that translates to say the microarray you talk about was used to make the 2 million probes.

As a complete aside, I'm a little surprised this isn't a Nature or Science paper.

Re:Labs out of business

[skids]

The hypothesized cost of $25 per sample was bandied about as feasible by the similar (same?) process VirScan.

I wish them luck. We may eventually be able to figure out just how widespread things like lyme disease really are.

Re:Doctor what's wrong with me?

[ColdWetDog]

In intact blood (well treated, processed correctly) the genomic DNA will be in cells which can be spun out quickly to create plasma. In fact, most lab tests are done on cell free fractions because the proteins and assorted other molecules bugger up the process.

You should also be able to determine in flanking sequences (if any). If you have a lot of extraneous DNA, it probably doesn't come from a free floating viral particle.

And finally, since the vast majority of viruses have some sort of protein capsid and are of a fairly constrained size, you can always fractionate the blood to include only those sized structures.

So, it is a potential issue but one that can be overcome in a fairly straightforward fashion.

Re:To much information not necessary a good thing.

[ColdWetDog]

Since we have very few antivirals, it isn't much of a problem.

"Go home and wash your hands. Don't kiss anybody you like." And that's pretty much it.

Re:Doctor what's wrong with me?

[Immerman]

I think resistance tends to be considerably more resilient than that in many/most cases. Influenza is actually one of the more volatile viruses we encounter on a regular basis - its actual mutation rate is much slower than something like HIV, but thanks to its incredibly widespread "host base" across the animal kingdom, and the facility with which it can "interbreed" with distant relatives, we are getting constantly barraged by radically new variants on a regular basis, a few of which manage to thrive. There's a reason the flu makes epidemiologists nervous - highly infectious, genetically unstable, and with a proven track record of occasionally spawning extremely lethal variants.

And of course, just for completeness sake, most of the things people commonly label as "the flu" are completely unrelated to influenza in the first place. Just part of the microbial medley passing through us on a regular basis.